Since the pioneering transformation advances of the early 1980's, much of the research efforts have been directed, and rightly so, to a horizontal spread of the technology. As a result of this emphasis, it is now possible to transform a wide variety of plant species. The trade off, however, has been less attention devoted to advancing the efficiency of the transformation process itself. Compared to many microbial systems, plant transformation appears somewhat antiquated. Whereas millions of independent transformants are routinely obtained with many microbial systems, in plants, the numbers are generally in the single to double-digit range. Hence a shotgun transformation approach to gene discovery is an option that has not been seriously entertained.
Unlike microbial gene transfer that requires analysis of relatively few representative clones due to the highly consistent phenotypes, plant gene transfer involves independent transformants that show highly variable levels and patterns of expression. Accordingly, for a typical DNA construct, twenty to fifty independent primary transformants are needed. For the commercial development of a new trait, hundreds of independent transformants are screened for the few with suitable transgene structure and expression.
The underlying reasons for the high variability in transgene expression in plants are not completely understood, but at least four factors are involved in this phenomenon. (1) Tissue culture: Somaclonal variation has long been associated with tissue culture regenerated plants. Changes in chromosome structure and ploidy, DNA sequence, DNA modification, and transposon activity have all been reported in somaclonal variants (Peschke and Phillips, 1992 Advances in Genetics, 30:41-75; Kaeppler et al., 2000 Plant Mol. Biol., 43:179-88). (2) Integration site: Chromosomal structures such as telomeres or heterochromatin are known to affect the expression of nearby genes (Stavenhagen and Zakian, 1994 Genes and Dev., 8:1411-22; Howe et al., 1995 Genetics, 140:1033-45; Wallrath and Elgin, 1995 Genes and Dev. 9:1263-77). As a transgene integrates at random locations, chromosomal influences on transgene expression can be expected to differ among independent transformants (Meyer, 2000 Plant Mol. Biol., 43:221-34). (3) Transgene redundancy: Transformed plants often contain variable numbers of transgenes. Rarely is there a positive correlation between gene expression and copy number. On the contrary, many cases have linked extra full or partial transgene copies to postrancriptional and transcriptional gene silencing (Muskens et al., 2000 Plant Mol. Biol., 43:243-60; Matzke et al., 2000 Plant Mol. Biol., 43:401-15). (4) Genetic mutations: As expected for any genetic manipulations, there is always the possibility of acquiring point mutations, deletions or rearrangements (Battacharyya et al., 1994 Plant J., 6:957-68).
Current methods in plant gene transfer often produce a complex integration structure at the insertion locus. Typically, multiple full and/or partial copies of the introduced molecule are arranged as direct and/or indirect repeats. Also inserted are selectable markers and other regulatory regions that are unnecessary after selection of a desired organism or plant containing the constructs. These complex patterns are not necessarily an impediment for research, but they are not desirable for commercial use. Structural documentation is a prerequisite for regulatory approval and a simple integration pattern is easier to characterize. Repetitive DNA also tends to be structurally and functionally unstable. Repeat sequences can participate in intra- and inter-chromosomal recombination. Even if a complex integration locus yields a suitable phenotype, it may be difficult to maintain the original structure, along with its defined expression pattern, through the numerous crosses involved in breeding and seed production programs. Multiple gene copies, particularly if some are arranged as indirect repeats, are frequently associated with homology-dependent gene silencing (Iyer et al., 2000 Plant Mol. Biol., 43:179-88; Muskens et al., 2000 supra).
Methods based on site-specific recombination systems have been described to obtain randomly integrated single copy transgenes by excising excess linked copies from the genome (Srivastava and Ow, 1999 Proc. Natl. Acad. Sci. USA, 96:11117-11121; Srivastava and Ow, 2001 Plant Mol. Biol. 46:561-566) and to insert DNA at a known chromosome location in the genome (O'Gorman et al., 1991 Science, 251:1351-55; Baubonis and Sauer, 1993 Nucl., Acids Res., 21:2025-29; Albert et al., 1995 Plant J., 7:649-59). These methods make use of site-specific recombination systems that are freely reversible. These reversible systems include the following: the Cre-lox system from bacteriophage P1 (Baubonis and Sauer, 1993, supra; Albert et al., 1995 Plant J., 7:649-59), the FLP-FRT system of Sacchromyces cerevisiae (O'Gorman et al., 1991, supra), the R-RS system of Zygosaccharomyces rouxii (Onouchi et al., 1995 Mol. Gen. Genet. 247: 653-660), a modified Gin-gix system from bacteriophage Mu (Maeser and Kahmann, 1991 Mol. Gen. Genet., 230:170-76), the β-recombinase-six system from a Bacillus subtilis plasmid (Diaz et al., 1999 J. Biol. Chem. 274: 6634-6640), and the γδ-res system from the bacterial transposon Tn1000 (Schwikardi and Dorge, 2000 FEBS let. 471: 147-150). Cre, FLP, R, Gin, β-recombinase and γδ are the recombinases, and lox, FRT, RS, gix, six and res the respective recombination sites (reviewed by Sadowski, 1993 FASEB J., 7:750-67; Ow and Medberry, 1995 Crit. Rev. Plant Sci. 14: 239-261).
The recombination systems above have in common the property that a single polypeptide recombinase catalyzes the recombination between two sites of identical or nearly identical sequences. Each recombination site consists of a short asymmetric spacer sequence where strand exchange takes place, flanked by an inverted repeat where recombinases bind. The asymmetry of the spacer sequence gives an orientation to the recombination site, and dictates the outcome of a recombination reaction. Recombination between directly or indirectly oriented sites in cis excises or inverts the intervening DNA, respectively. Recombination between sites in trans causes a reciprocal translocation of two linear DNA molecules, or co-integration if at least one of the two molecules is circular. Since the product-sites generated by recombination are themselves substrates for subsequent recombination, the reaction is freely reversible. In practice, however, excision is essentially irreversible because the probability of an intramolecular interaction, where the two recombination-sites are closely linked, is much higher than an intermolecular interaction between unlinked sites. The corollary is that the DNA molecule inserted into a genomic recombination site will readily excise out.
In contrast to the freely reversible recombination systems, there are recombination systems that can catalyze irreversible reactions. In one such system from bacteriophage phage λ, the λ integrase recombines non-similar sequences known as attB and attP to from attL and attR, respectively. This reaction requires DNA supercoiling of the target sites, and accessory proteins IHF and FIS. The reverse reaction, from attL×attR to form attB and attP, requires an additional excision-specific protein known as XIS. Mutant integrase proteins can perform intramolecular, but not intermolecular, reactions without these requirements. Using these mutant λ integrases, Lorbach et al. (2000 J. Mol. Biol., 296:1175-81) demonstrated DNA inversions in recombination targets introduced into the human genome.
A more useful irreversible recombination system described in the prior art is the Streptomyces phage φC31 recombination system. A 68 kDa integrase protein recombines an attB site with an attP site. These sites share only three base pairs of homology at the point of cross-over. This homology is flanked by inverted repeats, presumably binding sites for the integrase protein. The minimal known functional size for both the φC31 attB and attP is approximately 30 to 40 base pairs. The efficacy of the φC31 integrase enzyme in recombining its cognate attachment sites was demonstrated in vitro and in vivo in recA mutant Escherichia coli (Thorpe & Smith, 1998 Proc. Nat'l. Acad. Sci. USA, 95: 5505-10). Unlike the phage λ system, the φC31 integration reaction is simple in that it does not require a host factor. Unlike the phage λ mutant integrase system, it is capable of intermolecular as well as intramolecular reactions.
Prior art that uses reversible recombination systems require complicated strategies to keep the DNA from excising or exchanging back out from the genome. What are needed in the art are compositions and methods for achieving stable site-specific integration of transgenes such that 1) the DNA molecule is introduced as a single copy; 2) the inserted DNA does not readily excise back out, 3) excess DNA associated with the gene integration process, but is no longer needed afterwards, can be removed, and/or 4) additional DNA can be appended to the existing site adjacent to the previously inserted DNA.